GTWY logo
    Moonshot logoMoonshot

    kimi-k2.7-code-highspeed

    moonshot/kimi-k2.7-code-highspeed

    ChatVisionFilesToolsSystem prompt

    Context

    262K

    Max output

    262K

    Input / 1M

    $1.90

    Output / 1M

    $8.00

    Cached / 1M

    $0.38

    Cutoff

    —

    About this model

    Kimi K2.6 is Moonshot AI's latest and most intelligent model, with stronger and more stable long-horizon code generation, significantly improved instruction following and self-correction. It features a native multimodal architecture supporting text, image, and video input, thinking and non-thinking modes, and both dialogue and agent tasks. Supports automatic context caching, ToolCalls, JSON Mode, Partial Mode, and internet search.

    Best suited for

    • Long-horizon agentic task execution and self-correction
    • Stable, high-performance code generation, debugging, and refactoring
    • Long-context workflows up to 256K tokens with deep reasoning
    • Multimodal workflows (text + image + video inputs)
    • Complex research, data analysis, and information synthesis
    • Tool-calling agents with JSON/structured outputs
    • Internet-search-augmented question answering

    Built-in tools

    Web searchGtwy web search

    Hosted by the gateway — enable them per request without wiring your own endpoint.

    Capabilities

    Vision

    Accepts images alongside text in the same message.

    Files

    Accepts file attachments — PDFs, transcripts, spreadsheets.

    Tools

    Native function calling, so agents can invoke your endpoints.

    System prompt

    Honours a dedicated system role, separate from the user turn.

    Supported parameters

    max_tokens

    The maximum number of tokens to generate for the chat completion. Maps to max_completion_tokens.

    tools

    Lists tool definitions or capabilities available to the model.

    tool_choice

    Decides whether to use tools or just the model for generating responses.

    response_type
    additional_stop_sequencesStop Sequences

    Up to 5 stop sequences that halt generation when fully matched. Maps to stop.

    stream

    Sends the response in real-time as it's being generated.